bwa mem don't set PROPER_PAIR flag in pair end align

[gada]

$ cat r1.fq
@1
GGGCTAAGAGACAGCAAATACACATGAACAGAAAGAAGAGGGCAAAGAAAAGGCTGACGG
+
CCCFFFFFHHFHHJIJIJIJJJJJJJJDIJJJJJIJJIJJIIGIJJJJHIHJJJJIHIHI

$ cat r2.fq
@1
TAGTTTAAACGAGATTGCCAGGACCGGGTATCATTCACCATTTTTCTTTTCGTTAACTTG
+
CCCFFFFFHHHHHJJJIIIHIIJGGIGIJEIJIFGGGIJJJGFIIIJIFFEF?CFFEDCB

$ bwa mem ../ref/ref.fa r1.fq r2.fq > mem.sam;cat mem.sam
@SQ SN:1 LN:100000
@PG ID:bwa PN:bwa VN:0.7.10-r1017-dirty CL:bwa mem ../ref/ref.fa r1.fq r2.fq
1 81 1 1741 60 60M = 1681 -120 CCGTCAGCCTTTTCTTTGCCCTCTTCTTTCTGTTCATGTGTATTTGCTGTCTCTTAGCCC IHIHIJJJJHIHJJJJIGIIJJIJJIJJJJJIDJJJJJJJJIJIJIJHHFHHFFFFFCCC NM:i:1 MD:Z:18A41 AS:i:55 XS:i:0
1 161 1 1681 60 60M = 1741 120 TAGTTTAAACGAGATTGCCAGGACCGGGTATCATTCACCATTTTTCTTTTCGTTAACTTG CCCFFFFFHHHHHJJJIIIHIIJGGIGIJEIJIFGGGIJJJGFIIIJIFFEF?CFFEDCB NM:i:1 MD:Z:21C38 AS:i:55 XS:i:0

$ bwa aln ../ref/ref.fa r1.fq >r1.sai
$ bwa aln ../ref/ref.fa r2.fq >r2.sai
$ bwa sampe -r "@RG\tID:foo\tSM:bar" ../ref/ref.fa r1.sai r2.sai r1.fq r2.fq |\
samtools view -bS - > aln.bam
$ samtools view aln.bam
1 83 1 1741 60 60M = 1681 -120 CCGTCAGCCTTTTCTTTGCCCTCTTCTTTCTGTTCATGTGTATTTGCTGTCTCTTAGCCC IHIHIJJJJHIHJJJJIGIIJJIJJIJJJJJIDJJJJJJJJIJIJIJHHFHHFFFFFCCC RG:Z:foo XT:A:U NM:i:1 SM:i:37 AM:i:37 X0:i:1 X1:i:0 XM:i:1 XO:i:0 XG:i:0 MD:Z:18A41
1 163 1 1681 60 60M = 1741 120 TAGTTTAAACGAGATTGCCAGGACCGGGTATCATTCACCATTTTTCTTTTCGTTAACTTG CCCFFFFFHHHHHJJJIIIHIIJGGIGIJEIJIFGGGIJJJGFIIIJIFFEF?CFFEDCB RG:Z:foo XT:A:U NM:i:1 SM:i:37 AM:i:37 X0:i:1 X1:i:0 XM:i:1 XO:i:0 XG:i:0 MD:Z:21C38

the problem is:
samtools mpileup can't output valid results when the PROPER_PAIR flag not set.

[dariober]

If you want mpileup to include these cases use the -A option (-A: count anomalous read pairs). However, the question of way bwa mem doesn't like this pair remains...

[gada]

thanks. got it. Maybe bwa mem just like paired reads overlapped? I'll have a try.

[dariober]

Maybe bwa mem just like paired reads overlapped?

Mmm.. I doubt it... Non overlapping pairs make perfect sense and I don't think such a bug could have sneaked into bwa mem undetected.

[gada]

$ bwa mem ../ref/ref.fa r1.fq r2.fq
[M::bwa_idx_load_from_disk] read 0 ALT contigs
@SQ SN:1 LN:100000
@PG ID:bwa PN:bwa VN:0.7.10-r1017-dirty CL:bwa mem ../ref/ref.fa r1.fq r2.fq
[M:rocess] read 2 sequences (182 bp)...
[M::mem_pestat] # candidate unique pairs for (FF, FR, RF, RR): (0, 1, 0, 0)
[M::mem_pestat] skip orientation FF as there are not enough pairs
[M::mem_pestat] skip orientation FR as there are not enough pairs
[M::mem_pestat] skip orientation RF as there are not enough pairs
[M::mem_pestat] skip orientation RR as there are not enough pairs
[M::mem_process_seqs] Processed 2 reads in 0.000 CPU sec, 0.000 real sec
1 97 1 1861 60 120M = 1979 180 CTTCATCTGCAGGTGTCTGACTTCCAGCAACTGCTGGCCTGTGCCAGGGTGCAAGCTGAGCACTGGAGTGGAGTTTTCCTGTGGAGAGGAGCCATGCCTAGAGTGGGATGGGCCATTGTT CCCCFFFFFHHFHHJIJIJIJJJJJJJJDIJJJJJIJJIJJIIGIJJJJHIHJJJJIHIHICCFFFFFHHFHHJIJIJIJJJJJJJJDIJJJJJIJJIJJIIGIJJJJHIHJJJJIHIHI NM:i:0 MD:Z:120 AS:i:120 XS:i:0
1 145 1 1979 60 62M = 1861 -180 TTCATCTTCTGGCCCCTGTTGTCTGCATGTAACTTAATACCACAACCAGGCATAGGGGAAAG BBBCDEFFC?FEFFIJIIIFGJJJIGGGFIJIEJIGIGGJIIHIIIJJJHHHHHFFFFFCCC NM:i:0 MD:Z:62 AS:i:62 XS:i:0
[main] Version: 0.7.10-r1017-dirty
[main] CMD: bwa mem ../ref/ref.fa r1.fq r2.fq
[main] Real time: 0.019 sec; CPU: 0.004 sec

$ ruby -e 'puts 97.to_s(2)'
1100001
ruby -e 'puts 145.to_s(2)'
10010001

nothing changed.